2.5. Sensory analysisThe tests were

2.5. Sensory analysis
The tests were carried out in a sensory laboratory under artificial actinic light, at 25 °C and with air circulation. The sensory panel consisted of eleven highly trained evaluators for coffee beverages. The espresso was prepared, weighing 7 g of the blend and the extraction was performed using a professional espresso machine (1 group Standard, Brasilia, Italy). The development of thefinal product included a quantitative descriptive analysis of the final functional coffee blend (96% roasted coffee, of which 70% Robusta and 30% Arabica, 3% cocoa, 2% coffee silverskin and 1% golden coffee) by Novadelta panel of judges, who rated the organoleptic characteristics of the product by applying a scale of 1 to 5. The sensory panel for quantitative descriptive analysis considered 15 parameters: color, aspect, odor, acidity, bitterness, saltiness, sweetness, metallic, astringent, iodoformflavor, fermented taste, body,flavor, aftertaste and persistence.
2.6. Ochratoxin A analysis
The analytical method for ochratoxin A in the functional coffee blend combined the HPLC separation with enhancedfluorescence detection (Shimadzu LC-10ADVP, Japan), as described byCastellanos-Onorio
et al. (2011). Ten grams of functional coffee blend was extracted for 30 min with 100 mL of 3% methanol/bicarbonate solution (20/80) at 60 °C during 50 min; 5 mL of thefiltered extract was diluted with 40 mL of PBS buffer and cleaned through an immunoaffinity column. The quantification limit is 1.9μgkg−1. This analysis was performed at a portuguese governmental laboratory (ASAE).

2.7. Statistical analysis
The various measured parameters were analyzed using a one-way ANOVA (P≤0.05), followed by a Tukey test for mean comparison (95% confidence level). The software STATISTICA forWindows Release
8.0—Copyright ©StatSoft, INC was used for the performed statistical analyses.

3. Results and discussion
3.1. New coffee blend characteristics
The selection of the roasting degree attended the factors that determine the quality of thefinal beverage. In this context the effectiveness of the antioxidant capacity of extraction of Robusta coffee is considerably higher than in Arabica and the comparable method of extraction through boiling water under high pressure is the most efficient (Budryn & Nebesny, 2008). Accordingly, the Robusta coffees were submitted to a medium–light roasting to preserve their antioxidant compounds, related to the content of total polyphenols, particularly chlorogenic acids. Moreover, the medium roasts are also indicated for the espresso coffees instead of dark and intense roasts, matching the profile of acceptance of consumers. The golden coffee was incorporated to overcome the lack of cocoa chlorogenic acids and the coffee silverskin, as well as to increase these compounds in thefinal product, since the roasting process partially destroys those antioxidant molecules. In fact, by including extracts of green coffee such problem might be override, as previously pointed by Suzuki, Kagawa, Ochiai, Tokimitsu, & Saito (2002), further using alcohol (Thom, 2007)oramixture of both (Madhava-Naidu, Sulochanamma, Sampathu, & Srinivas, 2008). Therefore, the composition of thefinal blend included golden coffee, in order to extract components from its micro particles to the beverage. The extraction with espresso machines also provided an intense extract due to the size of the particles, the temperature and the water pressure.

3.2. Chemical analysis
3.2.1. Caffeine, trigonelline and theobromine
Taking into account the weight of the components of the functional coffee blend, caffeine did not strongly change with the roasting process, whereas trigonelline decreased to about half of the values observed in the green coffee bean (Bicho et al., 2013a). The caffeine content was lower in thefinal functional coffee blend relative to the commercial coffee blends in capsules and the coffee blends in a sealed package with a one-way degassing-valve (Table 2). Caffeine content was higher in the commercial coffee blends in capsules (probably linked to the smaller particle size that improved extraction) and in the coffee blend in a sealed package with a one-way degassing-valve due to the higher content of Robusta. Yet, caffeine extraction in the espresso beverages reached higher values in the functional coffee blend and was more efficient using the DQOOL machine relative to the Briel equipment (Table 3). The caffeine values obtained for our blend ranged between 2.96 and 3.50 mg mL −1 , being higher than those reported by Maeztu et al. (2001), that obtained 2.09 mg mL−1 for Arabica and 2.88 mg mL
−1 for Robusta espresso coffees, but were similar to the range 2.6 to 3.8 mg mL−1 reported byIlly & Vianni (1995). Also, the caffeine levels (standardized to 25 mL) in our blends, varied between 74 and 87.5 mg per cup, were close to the values reported byCasal, Alves, Eulália, & Oliveira (2009)that reported 54–92 mg for a medium espresso coffee of 30 mL cup. It must be pointed that all of our blend formulations corresponded to thefine grinding (milling machine), although a larger proportion of bigger particles was found in the sample of the coffee blend in a sealed package with a one-way degassing-valve and smaller particle size was observed in the commercial coffee blend in capsules (Table 4).
The coffee blends in a sealed package with one-way degassing-valve and the commercial coffee blend in capsules showed a similar trigonelline content when compared to the functional coffee blend
(Table 2). The extraction of trigonelline was similar in both coffee machines, as observed through the results of the functional coffee blend (Table 3). The content of trigonelline was significantly higher in the
functional coffee blend only when compared to the coffee blend in a sealed package with a one-way degassing-valve. The obtained values (between 0.91 and 1.10 mg mL−1) were close to the 1.15 mg mL
−1 found for espresso coffees (Maeztu et al., 2001). Note that, the theobromine content in defatted cocoa powder is usually around 15.5 mg g−1 (Belščak, Komes, Horžić,Ganić,&Karlović, 2009), but the partially defatted cocoa powder used in our blends presented a much higher content (44.2 mg g−1)(Table 5). The caffeine/ theobromine ratio in the functional coffee blend powder was 41:1 (calculated from values inTables 2 and 5), and was likely influenced by the high content of theobromine present in the cocoa used in that blend. Considering the theobromine extraction capability of both tested machines, the beverage obtained from the Briel system presented a somewhat higher content, visually related to the extraction of cocoa through the pores of thefilter holders that is more efficient than in DQOOL (where the retention of cocoa in thefilter pulp is higher).